ST3236: Stochastic Process Tutorial 9 TA: Mar Choong Hock Email: g0301492@nus.edu.sg Exercises: 10 Question 1 Messages arrive at a telegraph office as a Poisson process with mean rate 3 messages per hour (a)what is the probability that no message arrive during the morning hours 8:00am to noon? (b) what is the distribution of the time at which the first afternoon message arrives? Question 1a P(X(12) - X(8) = 0) = e-4 = e-12 Note that only the duration of the time is required, property of stationary process. Is Poisson Process a Markov Process? Question 1b Let U1 be the time for the first afternoon message P(U1 t) = P(X(t) - X(12) = 0) = e-(t-12); t > 12 Thus FU1(t) = 1 - e-(t-12); t > 12 • The time distribution is exponentially distributed with mean time = 1/. • In general, the inter-arrival time between two events for Poisson process is exponentially distributed, F(Vk) = 1 - e-(Vk) (k-1)th arrival Vk kth arrival t Question 2 Suppose that customers arrive at a facilities according to a Poisson process having rate = 2. Let X(t) be the number of customers that have arrived up to time t. Determine the following probabilities and conditional probabilities and expectations (a) P(X(1) = 2,X(3) = 6) (b) P(X(1) = 2|X(3) = 6) (c) P(X(3) = 6|X(1) = 2) (d) E{X(1)X(5)[X(3) - X(2)]} Question 2a Let = 2. Make use of the properties of the counting process and the independent of disjoint time intervals. Question 2b P X (1) 2, X (3) 6 P X (1) 2 X (3) 6 P( X (3) 6) 64 6 e 3 e 3 3 6 6! 64 6 e 3 e 6 6 6 6! 2 6 6 . 5 .4 .3 .2 3 66 160 243 Question 2c P X (1) 2, X (3) 6 P X (3) 6 X (1) 2 P ( X (1) 2) 64 6 e 3 e 2 2! 64 6 e 3 e 2 2 2 2 6 1 4 32 4 e e 2! 3 2 3 Or by counting property, P(X(3)=6|X(1)=2) = P(X(3)-X(1)=4) Question 2d t 1 2 3 5 Question 3 Let X(t) be a Poisson process of rate = 3 per hour. Find the conditional probability that there are two events in the first hour, given that there are five events in the first three hours. Question 3 Use Theorem in Lecture notes, Since we know there are 5 customers in the three hour interval, we want to know probability of any 2 customers in the 1-hour interval (and 3 customers in the 2-interval). Note that where the customers are distributed uniformly over the time interval, even though the arrival process is Poisson. Question 3 From Bayes’ Theorem, P(X(1)=2|X(3)=5) P(X(3)=5) = P(X(3)=5|X(1)=2) P(X(1)=2) = But, by counting property, P(X(3)=5|X(1)=2)= P(X(3)-X(1) = 3) Question 4 Customers arrive at a service facility according to a poisson process of rate customers/hour. Let X(t) be the number of customers that have arrived up to time t. Let W1,W2, … be the successive arrival times of the customers. Determine the conditional means E[W5 | X(t) = 4] and E[W3 | X(t) = 4] Question 4 Case E[W5 | X(t) = 4]? In this case, we know that up to time t, the 4th customer has arrived, what is the average waiting time s for the 5 customers to arrive? Question 4 Question 4 Average waiting time for five customer = average waiting time for four customer + average interarrival time between the fourth and the fifth customer. (See Q1b) Question 4 - Optional W5 V5 W4 1 2 3 4 t 5 E[W5 | X(t) = 4] = E[V5+W4 | W4 = t] = E[V5+t] = E[V5] + t =… Can try this on the lecture notes problem and split intervals to W3 + V4 + V5 Question 4 Case E[W3 | X(t) = 4]? Note: To say the waiting time for 3 customer is less than s is same as at time s, there is 3 or more customers arrivals. Because, if at time s there is 3, then s = W3, if more than 3, s > W3. Question 4 Case E[W3 | X(t) = 4]? Note: To say the waiting time for 3 customer is less than s is same as at time s, there is 3 or more customers arrivals. Because, if at time s there is 3, then s = W3, if more than 3, s > W3. Question 4 Question 5 Let X1(t) and X2(t) be independent Poisson process having parameters 1 and 2 respectively. What is the probability that X1(t) = 1 before X2(t) = 1? That is , what is the probability that first arrival for process 1 happened earlier than the first arrival for process 2. Question 5 Let W1 and W’1 be the waiting time for the first event in Poisson process 1 and 2 respectively. Then W1 and W’1 follow exponential distributions with parameters 1/1 and 1/2 respectively and are independent of each other. The probability is Question 5 PW1 W '1 P W1 W '1 W '1 t p (t )dt 0 P W1 W '1 W '1 t 2 e 2t dt 0 t 1s 1e ds 2 e 2t dt 0 0 1 e e 1t 2 2t dt 0 1 2 e 1 2 t dt 0 1 2 1 2 1 1 2